[YLUG] Data recovery End of the story ?

[Roger Leigh]

On Tue, Jul 28, 2009 at 04:58:40PM +0100, Patrick Dupre wrote:
>> On Tue, Jul 28, 2009 at 03:22:42PM +0100, Patrick Dupre wrote:
>>> On Tue, 28 Jul 2009, Roger Leigh wrote:
>>>
>>>> On Tue, Jul 28, 2009 at 12:20:23PM +0100, Patrick Dupre wrote:
>>>>
>>>>> 1) dd if=/dev/sdx of=/dev/sdy   did not work (incomplet data copy)
>>>>
>>>> Was this because the destination was too small, or an error
>>>> prevented a full copy?
>>> No, sdx was 80Go, sdy was 160 !B
>>>>
>>>>> 2) I was rescued by ddrescue /dev/sdb11 /dev/sdcx
>>>>
>>>> If dd failed while ddrescue worked, this implies that disk errors
>>>> caused the copy to abort.  ddrescue continues on error so that
>>>> it copies every /readable/ block, but if particular blocks are bad
>>>> they won't get copied.  The implication of this is that your disk
>>>> isn't just corrupted, it's broken and needs replacing.
>>>
>>> I do not think so, the partition table was just a bit messy.
>>> and the recover was perfect after a fsck -y
>>> the partition was in lost+found dir with a # number.
>>> The rest seems perfectly OK.
>>
>> dd copies raw blocks from your disk.  Whether the partition table was
>> messy or not is most likely irrelevant.  The only circumstance where
>> it could fail is if the partition extends over the upper size limit
>> of the disc (in which case you'll still have as full a backup as is
>> practical).  If this was not the case, then your disc is not well;
>> fsck might not have detected silent corruption of your data (as
>> opposed to the filesystem metadata).  If anything ended up in
>> lost+found, then /something/ was wrong.
>
> What can I tell you ?
> Do you want to test the drive ????

I don't want to test it personally, but I think that it would be
prudent for you to do so given that there's a possibility that
it has bad blocks or is otherwise defective.

As I said in a previous email, you can use smartctl (smartmontools)
to check the HDD SMART status.  This will let you see if the hard
disc controller thinks it's damaged (but this is not definitive).

You can also check whether the hard disc can physically read the disc
with badblocks.  Just run "badblocks /dev/sdx".  This tests if you
can read the whole disc without error.  As I mentioned in an earlier
email, there's a -w option to test that you can also write without
error as well (but only do this once you're certain you no longer
have important data on the disc).

If it were me, I'd do both of those at a minimum.


Regards,
Roger

-- 
  .''`.  Roger Leigh
 : :' :  Debian GNU/Linux             http://people.debian.org/~rleigh/
 `. `'   Printing on GNU/Linux?       http://gutenprint.sourceforge.net/
   `-    GPG Public Key: 0x25BFB848   Please GPG sign your mail.